Clinical Section: Research Article

Gerontology 2022;68:999–1009 Received: February 15, 2021

Accepted: October 14, 2021
DOI: 10.1159/000520318 Published online: January 4, 2022

Sex Differences in Cognitive Functioning

with Aging in the Netherlands
Astrid C.J. Nooyens a Hanneke A.H. Wijnhoven b Laura S. Schaap b
     

Lena D. Sialino a, b Almar A.L. Kok c Marjolein Visser b

     

W.M. Monique Verschuren a, d H. Susan J. Picavet a Sandra H. van Oostrom a

     

aCentre for Nutrition, Prevention and Health Services, National Institute for Public Health and the Environment,
Bilthoven, The Netherlands; bDepartment of Health Sciences, Faculty of Science, Amsterdam Public Health
Research Institute, Vrije Universiteit, Amsterdam, The Netherlands; cAmsterdam UMC, Vrije Universiteit Amsterdam,
Department of Epidemiology & Biostatistics, Amsterdam Public Health, Amsterdam, The Netherlands;
dJulius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University, Utrecht,

The Netherlands

Keywords cognitive function than men (p’s < 0.01). However, women
Aging · Cognitive function · Cohort effect · Longitudinal showed up to 10% faster decline in these cognitive domains,
study · Sex except for flexibility, where women showed 9% slower de-
cline. In the LASA, women scored poorer on fluid intelligence
(p < 0.01), but their decline was 10% slower than that in men.
Abstract Female advantage was larger in later born cohorts; adjust-
Introduction: Dementia prevalence in older women is high- ment for the educational level increased the female advan-
er than that in men. The purpose of the present study was to tage. Conclusion: Women have better memory and process-
investigate whether there is a female disadvantage in cogni- ing speed than men at middle age. This female advantage
tive functioning at adult age and/or whether a female disad- becomes smaller with aging and has increased in more re-
vantage develops with age. Methods: Data of 5,135 women cent birth cohorts. © 2022 The Author(s).
and 4,756 men from the Longitudinal Aging Study Amster- Published by S. Karger AG, Basel

dam (LASA) and the Doetinchem Cohort Study (DCS) were

used. In the LASA, memory, processing speed, fluid intelli-
gence, and global cognitive function were measured every Introduction
3–4 years since 1992 in persons aged 55+ years for up to 23
years. In the DCS, memory, processing speed, cognitive flex- In 2016, in the Netherlands, the number of women
ibility, and global cognitive function were measured every 5 who died with dementia was twice the number of men
years since 1995 in persons aged 45+ years for up to 20 years. who died with dementia [1]. Also, the prevalence of de-
Sex differences in cognitive aging were analyzed using linear mentia is almost twice as high in older women than in
mixed models and also examined by the 10-year birth cohort men of the same age [2]. Dementia is usually preceded by
or level of education. Results: Women had a better memory, a long period of decline in cognitive functions [3]. Based
processing speed, flexibility, and, in the DCS only, global on a continuum from healthy cognitive function to de-

[email protected] © 2022 The Author(s). Correspondence to:

www.karger.com/ger Published by S. Karger AG, Basel Astrid C.J. Nooyens, astrid.nooyens @ rivm.nl
This is an Open Access article licensed under the Creative Commons
Attribution-NonCommercial-4.0 International License (CC BY-NC)
(http://www.karger.com/Services/OpenAccessLicense), applicable to
the online version of the article only. Usage and distribution for com-
mercial purposes requires written permission.
mentia, a lower cognitive function (cognitive reserve) Medical Ethics Committee of the Netherlands Organization for
and/or a steeper cognitive decline increases the risk of Applied Scientific Research and the Medical Ethics Committee
of the University of Utrecht. All participants gave written in-
dementia. The purpose of the present study was to inves- formed consent.
tigate whether there is a female disadvantage in cognitive The LASA started in 1992 with 3,107 participants aged 54–85
functioning at adult age and/or whether a female disad- years from 3 birth cohorts (1908–1917, 1918–1927, and 1928–
vantage develops with age that can explain the sex differ- 1937). Ten years later, 1,002 persons aged 54–65 years (born be-
ences in dementia prevalence at older age. tween 1938 and 1947) were added, and in 2012, 1,023 persons aged
54–65 years (born between 1948 and 1957) were added to the ongo-
Sex differences in cognitive function at adult age are ing study. Participants are re-measured every 3–4 years. From the
well documented: women score higher on cognitive tasks LASA, 2,653 women and 2,469 men aged 54–85 years at baseline
that require verbal processing, while men score higher on could be included in our study (see online suppl. Table 1; for all
tasks that require visuospatial processing [4]. A system- online suppl. material, see www.karger.com/doi/10.1159/520318).
atic review showed that sex differences remain similar be- The median follow-up was 6 years and ranged from 0 to 23 years
(1–8 repeated measurements).
tween the age of 60–80 years, but that sex differences in The DCS started in 1987–1991 with an age- and sex-stratified
cognitive decline may occur after the age of 80 years, al- sample of 12,404 men and women aged 20–59 years from the town
though the directions of the associations found were of Doetinchem. A random selection of 7,769 of these participants
sometimes contradictory [5]. Since this review, some was re-invited 6 years later and subsequently has been invited ev-
more longitudinal studies showed inconsistent results on ery 5 years. Halfway the second examination round (from 1995
onward), cognitive testing was introduced for participants aged
sex differences in cognitive decline: either no difference 45 years and over. From the DCS, 2,482 women and 2,287 men
in the rate of cognitive decline between men and women aged 45–82 years at baseline, born between 1928 and 1967, were
[6], or a steeper decline for men than for women [7, 8]. included (see online suppl. Table 2). The median follow-up was
The longest follow-up in these studies was 17 years [5–8]. 10 years and ranged from 0 to 20 years (1–5 repeated measure-
In a recent US study with up to 21 years of follow-up, ments).
women had faster declines in global function and execu- Cognitive Function
tive function, but there was no difference in memory de- In the LASA 4 cognitive tests have been repeatedly carried out:
cline between men and women [9]. In this latter study, a 15-word learning test [17], a letter substitution task [18], the Ra-
only few repeated measurements were available for mem- ven Colored Progressive Matrices [19] (up to and including the 6th
ory. In the present study, we were able to examine cogni- measurement round), and the Mini-Mental State Examination
(MMSE) [20]. In the DCS, 4 cognitive tests are repeatedly carried
tive change over a longer period of time (up to 23 years), out: a 15-word learning test [21], a letter digit substitution task
based on a higher number of repeated cognitive measure- [22], the Stroop test [23], and a fluency test [24]. Details on these
ments and up to a higher age (100 years), in 2 large Dutch cognitive tests are included as online supplementary material.
cohort studies. Since secular trends in economic prosper-
ity and gender equity in educational chances may influ- Cognitive Domain Scores
In the LASA, memory function was calculated as the average of
ence sex differences in cognitive function [4, 10], we also the total score on the immediate recalls, the maximum score on the
studied whether sex differences in cognitive functioning 3 immediate recalls, and the score on the delayed recall of the 15-
differ between different birth cohorts and between differ- word learning test. Information processing speed was calculated as
ent levels of education. We expect that secular trends have the average score on the 3 trials of the letter substitution test. Fluid
improved the average cognitive function of women more intelligence was calculated as the total score on sections A and B of
the Raven Progressive Matrices. Global cognitive function was the
than that of men, as the educational opportunities of maximum score on the MMSE, either based on the subtraction or
women have improved in the course of the 20th century the spelling backward item.
due to women’s emancipation [11] and education is a key In the DCS, memory function was calculated as the average of
predictor of later life cognitive functioning. the total score on the immediate recalls, the maximum score on the
3 immediate recalls, and the score on the delayed recall of the 15-
word learning test. Information processing speed was calculated as
the average of the first 2 trials of the Stroop test (reading and color
Methods naming) and the score on the letter digit substitution task. Flexibil-
ity was calculated as the score on the third trial of the Stroop test
We used data from 2 large Dutch aging cohorts: the Longitu- (naming ink color). All tests are taken together to calculate a mea-
dinal Aging Study Amsterdam (LASA) [12–14] and the sure for global cognitive function (average z-score on the total of
Doetinchem Cohort Study (DCS) [15, 16]. Both studies have immediate recalls, delayed recall, third trial of the Stroop test
been conducted in line with the Declaration of Helsinki. The (naming ink color), letter digit substitution test, and the fluency
LASA has been approved by the Medical Ethics Committee of test) [25].
the VU University medical center and the DCS by the external

1000 Gerontology 2022;68:999–1009 Nooyens et al.

DOI: 10.1159/000520318
 Demographic Characteristics Table 1. Baseline characteristics (participants at their first cognitive
For all participants in both the LASA and DCS, date of birth, measurement)
age, sex, and the highest attained level of education are ascertained.
Based on birth date, participants were categorized in 10-year birth Women Men
cohorts: 1908–1917, 1918–1927, 1928–1937, 1938–1947, 1948–
1957, or 1958–1967. The highest attained level of education was LASA n = 2,654 (52%) n = 2,469
categorized in 3 levels, based on the distribution in LASA: low (el- Mean age (SD), years 66.5 (8.7) 66.7 (8.8)
ementary education or less), medium (lower vocational education Range 54–88 54–88
or general intermediate education), and high (intermediate voca- Level of education, %
tional education, general secondary education, or higher). Low* 39.8 25.8
Medium 32.6 31.3
Statistics High 27.6 42.9
In order to compare cognitive function between men and Birth cohort, %
women, reducing the impact of learning effects on cognitive tests, 1908–1917 21.6 23.5
all cognitive data were corrected for the number of cognitive mea- 1918–1927 19.3 18.3
surements so far, based on the association between the number of 1928–1937 19.4 19.1
measurements (classes) and cognitive function in a linear longitu- 1938–1947 19.8 19.1
dinal regression model adjusting for age, age2, sex, and level of 1948–1957 19.9 20.1
education. After adjusting for the number of measurements, all
DCS n = 2,482 (52%) n = 2,287
cognitive data, except for the MMSE scores, were standardized
Mean age (SD), years 55.2 (6.6) 55.5 (6.6)
into z-scores based on the mean and standard deviation of the first
Range 45–77 45–82
cognitive measurements of all participants in the study (baseline).
Level of education, %
Low* 6.7 5.7
Sex Differences
Medium 48.2 31.9
To study sex differences in (mean level of) cognitive function,
High 45.1 62.4
cognitive test scores were included in mixed model analyses with
Birth cohort, %
random intercept for the individuals, random slope for age (since
1928–1937 17.6 19.1
this increased model fit significantly, as tested by comparing -2LL
1938–1947 27.9 31.1
between model without and with random slope for age), and an
1948–1957 37.7 36.5
unstructured covariance matrix. Independent variables in these 1958–1967 16.9 13.3
analyses on differences in the level between women and men were
sex, age, and age2 (model 1). If the estimate for sex is statistically Baseline characteristics are given over all first measurements
significant in this model, positive values denote that women per- (t0). LASA, Longitudinal Aging Study Amsterdam; DCS, Doetinchem
form better than men do. Negative values denote that men perform Cohort Study. * Level of education was classified in 3 levels: Low,
better than women do. elementary education or less; Medium, lower vocational education
Based on this model, difference in cognitive age (expressed in or general intermediate education; High, intermediate vocational
years) was calculated in order to facilitate the interpretation of the education, general secondary education, or higher (higher
differences between men and women at a certain calendar age. At vocational education/college education/university education).
what calendar age is the cognitive function of men at the same
level as the cognitive function of women at the age of 65 years? The
difference in calendar age is used as an indicator for the difference
in cognitive age between men and women (see also online suppl. Otherwise, model 1 was used. If model 2 was used, so decline with
Fig. 1). So if the cognitive function in women at age 65 years equals aging differed between men and women, it was then examined at
the cognitive function of men at age 62 years, the difference in cog- which age the sex differences were no longer statistically signifi-
nitive age is 3 years [26]. cantly different.
To test for sex differences in cognitive decline with aging, the Associations were considered to be statistically significant at
interaction term of sex and age was added. Independent variables p < 0.05. Age was centered at 55 years in all analyses on sex differ-
in these analyses on differences in cognitive decline with aging ences, in order to present a realistic sex difference (at age 55 years),
were sex, age, age2, sex × age, and sex × age2 (model 2). If an inter- especially when cognitive decline with aging differs between men
action term of sex and age is significant in this model, positive val- and women.
ues denote that the decline in cognitive function with aging is
stronger in men than that in women. Negative values denote that Differences between Birth Cohorts and Levels of Education
the decline with aging is stronger in women than that in men. Sex Sex differences in level of cognitive function were tested between
× age2 was only included if this interaction term was significantly birth cohorts or between educational levels by adding an interac-
associated with cognitive function at p < 0.05. tion term of sex and the subgroups to model 1 (e.g., sex*education).
Based on these models, difference in cognitive aging was calcu- Sex differences in cognitive decline with aging between subgroups
lated as difference in cognitive decline (in percentage) between were tested by adding an interaction term of sex, age, and the sub-
men and women between ages 65–75 years. If the interaction term groups (3-way interaction: e.g., age*sex*education) to model 2. Sex
of sex and age was significant in model 2, this model was used to differences in cognitive function and cognitive decline with aging
describe and plot sex differences in cognitive function with aging. within subgroups were calculated in subgroup-stratified analyses.

Sex Differences in Cognitive Aging Gerontology 2022;68:999–1009 1001

DOI: 10.1159/000520318
 Table 2. Differences in the level of cognitive functioning and cognitive decline with aging between men and women, from middle to old age (range 55–104 years), and by

1002
birth cohort and level of education in LASA

Memory function Information processing speed Fluid intelligence Global cognitive function
(z-score) (z-score) (z-score) (MMSE score)

model 1 (level of model 2 (cognitive model 1 (level of model 2 (cognitive model 1 (level of model 2 (cognitive model 1 (level of model 2 (cognitive
cognitive functioning) decline) cognitive functioning) decline) cognitive functioning) decline) cognitive functioning) decline)

Overall (n = 4,632, 14,411 observations) (n = 4,579, 14,236 observations) (n = 3,860, 12,041 observations) (n = 5,123, 17,319 observations)
Sex† 0.422 [0.375, 0.469] 0.483 [0.409, 0.556] 0.106 [0.058, 0.154] 0.185 [0.122, 0.248] −0.143 [−0.194, −0.092] −0.216 [−0.296, −0.136] 0.002 [−0.110, 0.114] 0.044 [−0.132, 0.219]
Sex × age −0.004 [−0.008, −0.000] −0.006 [−0.009, −0.003] 0.005 [0.001, 0.009] −0.004 [−0.017, 0.009]

By birth cohort
1908–1917
Sex 0.30 [0.20, 0.40]a,b 0.37 [−0.08, 0.81] −0.03 [−0.14, 0.07]d −0.00 [−0.35, 0.35] −0.12 [−0.23, −0.01] −0.02 [−0.49, 0.45] −0.14 [−0.56, 0.29] 0.49 [−1.49, 2.46]
Sex × age −0.00 [−0.02, 0.01] −0.00 [−0.01, 0.01] −0.00 [−0.02, 0.01] −0.03 [−0.10, 0.05]j

DOI: 10.1159/000520318
1918–1927
Sex 0.47 [0.37, 0.58]a 0.60 [0.37, 0.82] 0.09 [−0.02, 0.20]e 0.13 [−0.06, 0.32] −0.03 [−0.14, 0.08]h,i −0.16 [−0.38 0.07] 0.05 [−0.22, 0.32] −0.09 [−0.78, 0.60]

Gerontology 2022;68:999–1009
Sex × age −0.01 [−0.02, 0.00] −0.00 [−0.01, 0.01] 0.01 [−0.00, 0.02] 0.01 [−0.03, 0.05]
1928–1937
Sex 0.44 [0.34, 0.54] 0.44 [0.32, 0.57] 0.07 [−0.04, 0.18]f 0.12 [−0.00, 0.24] −0.20 [−0.29, −0.10]h −0.21 [−0.33, −0.09] −0.09 [−0.30, 0.13] −0.16 [−0.44, 0.12]
Sex × age −0.00 [−0.01, 0.01] −0.00 [−0.01, 0.00] 0.00 [−0.01, 0.01] 0.01 [−0.01, 0.03]j
1938–1947
Sex 0.36 [0.26, 0.46]c 0.39 [0.26, 0.52] 0.11 [0.00, 0.21]g 0.18 [0.07, 0.30] −0.22 [−0.31, −0.12]i −0.25 [−0.39, −0.11] −0.07 [−0.32, 0.17] −0.04 [−0.35, 0.27]
Sex × age −0.00 [−0.01, 0.01] −0.01 [−0.02, −0.00] 0.01 [−0.01, 0.02] −0.00 [−0.03, 0.02]
1948–1957
Sex 0.55 [0.44, 0.66]b,c 0.54 [0.33, 0.75] 0.30 [0.20, 0.40]d,e,f,g 0.24 [0.09, 0.39] ‡ 0.19 [−0.02, 0.40] 0.42 [0.01, 0.83]
Sex × age 0.00 [−0.02, 0.03] 0.01 [−0.01, 0.03] −0.03 [−0.09, 0.02]

By educational level
Low
Sex 0.45 [0.36, 0.53] 0.61 [0.45, 0.76] 0.15 [0.07, 0.24]k 0.32 [0.18, 0.45] 0.03 [−0.06, 0.12] 0.12 [−0.06, 0.30] 0.32 [0.03, 0.60] 0.53 [0.02, 1.03]
Sex × age −0.01 [−0.02, −0.00] −0.01 [−0.01, −0.00] −0.00 [−0.01, 0.00]m −0.01 [−0.04, 0.01]n
Medium
Sex 0.55 [0.48, 0.63] 0.40 [0.24, 0.56] 0.34 [0.27, 0.42]k,l 0.39 [0.29, 0.49] 0.01 [−0.07, 0.09] −0.15 [−0.27, −0.02] 0.26 [0.09, 0.42] 0.04 [−0.22, 0.31]
Sex × age 0.03 [0.01, 0.04] −0.00 [−0.01, 0.00] 0.01 [0.00, 0.02]m 0.02 [0.00, 0.04]n
Sex × age2 −0.00 [−0.00, −0.00]
High
Sex 0.54 [0.46, 0.61] 0.60 [0.49, 0.71] 0.18 [0.11, 0.26]l 0.25 [0.16, 0.34] −0.07 [−0.14, −0.00] −0.14 [−0.24, −0.03] 0.12 [−0.00, 0.24] 0.09 [−0.13, 0.30]

Nooyens et al.
Sex × age −0.01 [−0.01, −0.00] −0.01 [−0.01, −0.00] 0.01 [−0.00, 0.01] 0.00 [−0.01, 0.02]

Estimates in bold are statistically significant at p < 0.05; estimates with same superscript letters are significantly different at p < 0.05 (between comparable models only). LASA, Longitudinal Aging Study Amsterdam; MMSE,
Mini-Mental State Examination. † Sex differences are presented as average [95% CL] and represent the average difference in cognitive function between men and women, regardless of age (model 1) or at age 55 years (model 2).
Positive values represent better scores (sex) or less decline for every year of aging (sex × age and sex × age2) for women compared to men. Sex differences are adjusted for age and age squared, with age centered at age 55 years.
‡ 
Fluid intelligence was not assessed in the latest birth cohort. For stratified analyses on the latest birth cohort (1948–1957), no random slope was included in the model.
 Sensitivity Analyses Cognitive Flexibility
In order to test whether menopause influenced results, analyses Women scored better than men on cognitive flexibil-
for DCS were repeated, excluding participants younger than 55
years. Mixed model analyses are able to deal with missing values, ity (Table 3). At the age of 65 years, women were cogni-
based on the assumption that missings are at random. Since miss- tively 3 years younger than men. This female advantage
ings due to attrition are likely not at random, we repeated our anal- further increased with aging due to smaller decline in
yses based on data up to age 80 years (LASA and DCS), diminish- women (9% between ages 65 and 75 years, see also Fig. 1).
ing attrition not at random due to, e.g., illness and death, and test-
ed whether sex differences were different between participants
who did and did not participate for 10 years (LASA) or in the most Global Cognitive Function
recent measurement round (DCS). In the LASA, no significant differences in global cogni-
tive function or decline in global cognitive function (based
on MMSE) with aging were observed between men and
Results women (Table 2; Fig. 1). In the DCS, women scored better
than men on global cognitive function (based on all cogni-
In the LASA, the average age at baseline was 66 years and tive tests, Table 3). At the age of 65 years, women were
in the DCS, 55 years. In both cohorts, there were about as cognitively 4 years younger than men. This female advan-
many men as women. In all birth cohorts, men were higher tage became smaller with aging due to stronger decline in
educated than women and each subsequent birth cohort women (10% between ages 65 and 75 years). Despite this
was higher educated than the previous one (see online sup- convergence, the sex difference in global cognitive func-
pl. Table 3). More details on demographic characteristics of tion in the DCS remained statistically significant until af-
the study population are presented in Table 1. ter the age of 85 years (see also Fig. 1).

Sex Differences in Cognitive Function Sex Differences by Birth Cohorts

Memory Female advantage was larger in the most recent birth
Women scored better than men on memory in the cohort than the earlier birth cohorts for memory and in-
LASA and DCS (Tables 2, 3). At the age of 65 years, wom- formation processing speed (Tables 2, 3), and global cog-
en were cognitively >10 years younger than men. This nitive function (in DCS only, Table 3). In analyses strati-
female advantage became smaller with aging due to stron- fied for birth cohort, cognitive decline was not consis-
ger decline in women (9% between ages 65–75 years). De- tently different between men and women, or between
spite this convergence, the sex difference in memory re- birth cohorts.
mained statistically significant until after the age of 85–90
years (see also Fig. 1). Sex Differences by Educational Levels
Sex differences in the level of cognitive function were
Information Processing Speed not statistically different between levels of education, ex-
In the LASA and DCS, women scored better than men cept for information processing speed in LASA (Tables 2,
on information processing speed (Tables 2, 3). At the age 3). For processing speed, the female advantage was sig-
of 65 years, women were cognitively 3 years younger than nificantly larger in the medium educated participants
men. This female advantage became smaller with aging than in the low- and high-educated groups. The female
due to stronger decline in women (7–10% between ages advantage in cognitive function within different levels of
65–75 years). Due to this convergence, the sex difference education appeared larger than the overall female advan-
in processing speed lost its statistical significance between tage. Moreover, sex differences in fluid intelligence were
the ages of 75–80 years (see also Fig. 1). no longer present, while for global cognitive function in
the LASA, a female advantage appeared in stratified anal-
Fluid Intelligence yses. Also, for fluid intelligence, cognitive flexibility, and
Women scored lower than men on fluid intelligence global cognitive function (LASA only), women in the me-
(Table 2). At the age of 65 years, women were cognitively dium educated group showed less cognitive decline than
4 years older than men. This female disadvantage became those in the low-educated group (Tables 2, 3).
smaller with aging due to smaller decline in women (10%
between ages 65 and 75 years). The sex difference in fluid Sensitivity Analyses
intelligence lost its statistical significance between the age Analyses excluding participants younger than 55 years
of 80–85 years (see also Fig. 1). showed similar or smaller estimates for overall sex differ-

Sex Differences in Cognitive Aging Gerontology 2022;68:999–1009 1003

DOI: 10.1159/000520318
 LASA DCS
29 1.0
0.5 Men
27

Gloabl cognitive function

Women
25 0
MMSE score

–0.5
23

(z-score)
–1.0
21
–1.5
19
–2.0
17 –2.5
15 –3.0
55 60 65 70 75 80 85 90 95 45 50 55 60 65 70 75 80 85
Age, years Age, years

1.0 1.0
Memory function (z-score)

Memory function (z-score)

0.5 0.5
0 0
–0.5 –0.5
–1.0 –1.0
–1.5 –1.5
–2.0 –2.0
–2.5 –2.5
–3.0 –3.0
55 60 65 70 75 80 85 90 95 45 50 55 60 65 70 75 80 85
Age, years Age, years

1.0 1.0
Information processing speed

Information processing speed

0.5 0.5
0 0
–0.5 –0.5
(z-score)

(z-score)

–1.0 –1.0
–1.5 –1.5
–2.0 –2.0
–2.5 –2.5
–3.0 –3.0
55 60 65 70 75 80 85 90 95 45 50 55 60 65 70 75 80 85
Age, years Age, years

1.0 1.0
Cognitive flexibility (z-score)

0.5 0.5
Fluid Intelligence (z-score)

0 0
–0.5 –0.5
–1.0 –1.0
–1.5 –1.5
–2.0 –2.0
–2.5 –2.5
–3.0 –3.0
55 60 65 70 75 80 85 90 95 45 50 55 60 65 70 75 80 85
Age, years Age, years

Fig. 1. Cognitive functions with aging for men and women in the LASA (plotted for age 55–99 years) and DCS
(plotted for age 45–85 years). Note that MMSE scores with aging of men and women in LASA are similar, and
therefore, only 1 line is visible in the figure. LASA, Longitudinal Aging Study Amsterdam; DCS, Doetinchem
Cohort Study; MMSE, Mini-Mental State Examination.

1004 Gerontology 2022;68:999–1009 Nooyens et al.

DOI: 10.1159/000520318
 Table 3. Differences in the level of cognitive functioning and cognitive decline with aging between men and women, from middle to old age (range 45–85 years) and by
birth cohort and level of education in DCS

Memory function Information processing speed Cognitive flexibility Global cognitive function
(z-score) (z-score) (z-score) (z-score)

model 1 (level of model 2 (cognitive model 1 (level of model 2 (cognitive model 1 (level of model 2 (cognitive model 1 (level of model 2 (cognitive
cognitive functioning) decline) cognitive functioning) decline) cognitive functioning) decline) cognitive functioning) decline)

Sex Differences in Cognitive Aging

Overall (n = 4,763, 12,883 observations) (n = 4,751, 12,834 observations) (n = 4,754, 12,838 observations) (n = 4,742, 12,756 observations)
Sex† 0.431 [0.382, 0.480] 0.447 [0.396, 0.499] 0.194 [0.143, 0.246] 0.209 [0.156, 0.261] 0.160 [0.110, 0.209] 0.149 [0.098, 0.200] 0.292 [0.242, 0.342] 0.317 [0.265, 0.368]
Sex × age −0.004 [−0.008, −0.000] −0.005 [−0.008, −0.002] −0.002 [−0.007, 0.003] −0.007 [−0.010, −0.004]
Sex × age2 0.000 [0.000, 0.001]

By birth cohort
1928–1937
Sex 0.24 [0.12, 0.35]e,f,g 0.35 [0.16, 0.54] −0.04 [−0.17, 0.09]i,j,k 0.35 [0.03, 0.68] 0.08 [−0.05, 0.20] 0.30 [−0.07, 0.67] 0.00 [−0.12, 0.12]a,b,c 0.09 [−0.07, 0.26]
Sex × age −0.01 [−0.02, 0.00] −0.05 [−0.09, −0.01] −0.04 [−0.08, 0.01] −0.01 [−0.02, 0.00]
Sex × age2 0.00 [0.00, 0.00] 0.00 [0.00, 0.00]
1938–1947
Sex 0.48 [0.39, 0.57]e 0.48 [0.37, 0.59] 0.21 [0.11, 0.30]i 0.21 [0.11, 0.31] 0.18 [0.09, 0.27] 0.15 [0.05, 0.25] 0.33 [0.24, 0.42]a 0.36 [0.26, 0.46]
Sex × age 0.01 [−0.01, 0.02] −0.00 [−0.01, 0.00]l 0.01 [−0.00, 0.01]m −0.01 [−0.01, −0.00]
Sex × age2 −0.00 [−0.00, −0.00]
1948–1957
Sex 0.43 [0.35, 0.50]f,h 0.43 [0.35, 0.50] 0.23 [0.15, 0.31]j 0.22 [0.14, 0.30] 0.15 [0.07, 0.23] 0.15 [0.07, 0.23] 0.31 [0.23, 0.39]b,d 0.30 [0.22, 0.38]
Sex × age 0.00 [−0.01, 0.01] −0.01 [−0.01, −0.00]l −0.00 [−0.01, 0.00]m −0.01 [−0.01, −0.00]
1958–1967
Sex 0.57 [0.43, 0.70]g,h 0.56 [0.42, 0.70] 0.32 [0.18, 0.45]k 0.32 [0.18, 0.46] 0.19 [0.07, 0.31] 0.18 [0.06, 0.31] 0.45 [0.32, 0.57]c,d 0.45 [0.32, 0.58]
Sex × age −0.00 [−0.04, 0.03] 0.00 [−0.02, 0.03] −0.00 [−0.02, 0.02] 0.00 [−0.03, 0.03]

DOI: 10.1159/000520318
By educational level

Gerontology 2022;68:999–1009
Low
Sex 0.52 [0.33, 0.72] 0.54 [0.29, 0.80] 0.30 [0.03, 0.56] 0.36 [0.06, 0.65] 0.38 [0.13, 0.63]q 0.45 [0.16, 0.74] 0.41 [0.21, 0.62] 0.56 [0.31, 0.80]
Sex × age −0.00 [−0.02, 0.01] −0.01 [−0.03, 0.01] −0.01 [−0.03, 0.01]r −0.02 [−0.03, −0.00]
Medium
Sex 0.48 [0.41, 0.56] 0.48 [0.40, 0.56] 0.32 [0.24, 0.40]p 0.34 [0.25, 0.42] 0.24 [0.16, 0.32] 0.22 [0.14, 0.30] 0.38 [0.30, 0.45] 0.40 [0.32, 0.48]
Sex × age −0.00 [−0.01, 0.01] −0.00 [−0.01, 0.00] 0.01 [0.00, 0.01]r −0.01 [−0.01, −0.00]
High
Sex 0.52 [0.45, 0.58] 0.54 [0.47, 0.60] 0.23 [0.17, 0.29]p 0.24 [0.18, 0.30] 0.20 [0.14, 0.26]q 0.19 [0.13, 0.25] 0.41 [0.35, 0.47] 0.43 [0.36, 0.49]
Sex × age −0.01 [−0.01, −0.00] −0.00 [−0.01, −0.00] −0.00 [−0.01, 0.00] −0.01 [−0.01, −0.00]
Sex × age2 0.00 [0.00, 0.00]

Estimates in bold are statistically significant at p < 0.05; estimates with same superscript letters are significantly different at p < 0.05 (between comparable models only). For stratified analyses on the latest birth cohort (1958–
1967), no random slope was included in the model. DCS, Doetinchem Cohort Study. † Sex differences are presented as average [95% CL] and represent the average difference in cognitive function between men and women,
regardless of age (model 1) or at the age of 55 years (model 2). Positive values represent better scores (sex) or less decline for every year of aging (sex × age and sex × age2) for women than men. Sex differences are adjusted for
age and age squared, with age centered at the age of 55 years.

1005
ences in cognitive function, but sex differences in cogni- disadvantage on fluid intelligence with more recent birth
tive change were larger for memory, and decline in flex- cohort is not easily explained. Based on our results, exist-
ibility was now significantly larger in men than that in ing differences between men and women seem to have
women (online suppl. Table 6). Analyses on data from enlarged over time.
participants up to 80 years old yielded comparable results In agreement with the systematic review [5], earlier
to those based on all available data (online suppl. Table analyses on sex differences in cognitive decline over, re-
7). In the LASA, sex differences in memory function, but spectively, 6 and 16 years of follow-up in the LASA re-
not memory decline, were larger in the group who re- vealed no significant sex differences [27, 28]. However, in
mained >10 years in the study than persons who dropped the present study with 23 years of follow-up, we did ob-
out earlier. In the DCS, although persons who did not serve sex differences in cognitive decline in the LASA.
participate in the most recent measurement round had The disagreement may be explained by the lack of sex dif-
worse cognitive function and declined faster, attrition did ferences in cognitive decline in earlier birth cohorts in the
not affect sex differences in cognitive function or cogni- LASA (Table 2). Later studies observed either no sex dif-
tive decline. ferences in cognitive decline [6], or, in contrast to our
results, observed less cognitive decline among women [7,
8]. For instance, in the English Longitudinal Study of
Discussion Ageing [8], women aged 50 years and over showed sig-
nificantly less decline in memory and global cognitive
The objective of this study was to investigate sex dif- function than men over 8-year period of follow-up. One
ferences in cognitive function at middle age and in cogni- recent study among US adults [9] showed faster cognitive
tive decline with aging from middle to old age. We ob- decline in women for global cognitive function and ex-
served a female advantage for memory function, infor- ecutive function, but not for memory. In both our co-
mation processing speed, and cognitive flexibility; a horts, cognitive decline in memory function as well as
female disadvantage for fluid intelligence; and no sex dif- processing speed was faster in women than in men. The
ferences (in LASA) or a female advantage (in DCS) for differences in results can possibly be explained by the fact
global cognitive function. Overall, these sex differences in that we have more repeated measurements, and thus
cognitive function declined with aging, since women more power, for memory than in Levine’s study.
showed faster cognitive decline in memory and informa- Memory is typically first affected in Alzheimer’s dis-
tion processing speed than men. Sex differences in cogni- ease, while processing speed is, among other domains,
tive function were larger in more recent birth cohorts. affected in vascular dementia. Therefore, the stronger de-
After adjustment for the educational level, the female ad- cline in these domains in women may indicate higher risk
vantages were larger, indicating that, particularly in ear- of these types of dementia at older age in women. We did
lier cohorts, the lower average education level of women not investigate sex differences in different types of de-
suppressed part of the sex differences in cognitive func- mentia, but questioned whether sex differences in cogni-
tioning. tive function and/or cognitive decline at adult age exist.
In our study, men outperformed women in spatial These may explain sex differences in the total prevalence
tests (fluid intelligence) and women outperformed men of dementia, based on the assumption that lower cogni-
on verbal tests (memory test) and speed tests (substitu- tive reserve and/or stronger cognitive decline in general
tion test). In an opinion article, Hyde [4] wrote that such increases the risk of dementia at older age.
differences between men and women existed in psycho- In our analyses, with up to 23 years of follow-up, we
logical research from the 1930s through the 1970s, but observed that, in general, sex differences in cognitive
that these phenomena may have narrowed over time [4]. functioning became smaller from middle into old age.
In our longitudinal datasets, however, we observed that Cognitive function was modeled as a (nonlinear) func-
both the female advantages (memory and information tion of age, since age is the most important determinant
processing speed) as well as the female disadvantage (flu- of cognitive functioning [5]. Sex significantly interacted
id intelligence) became larger in later birth cohorts. The with age on cognitive functioning in the overall analyses.
larger female advantages in more recent birth cohorts In the stratified analyses, this interaction was less clear
may be explained by the fact that inequalities in educa- and significant in only few birth cohorts (Tables 2, 3). It
tional attainment between men and women declined was not clear whether the sex differences in cognitive de-
across birth cohorts. However, the increase of the female cline in the overall models actually were artifacts based on

1006 Gerontology 2022;68:999–1009 Nooyens et al.

DOI: 10.1159/000520318
differences in sex differences between birth cohorts (ear- Especially in earlier birth cohorts, women are lower edu-
lier cohorts showing smaller sex differences in level of cated, and may therefore have lower cognitive reserve
cognitive function than later born cohorts). Therefore, ad than men. However, women score higher on most cogni-
hoc analyses were performed with birth cohort and edu- tive tests, which is not in line with this explanation. Third,
cational level as covariates in the models. Now, again, sig- selective survival or competing risk could explain sex dif-
nificant interactions of sex and age on cognitive function- ferences in cognitive decline. Based on the results in the
ing were observed (online suppl. Tables 4, 5), indicating Framingham study [33], it is questioned whether the
that the sex differences in cognitive decline in our popu- higher lifetime risk of dementia for women actually is a
lation were not only artifacts but also present within birth female disadvantage. An important competing risk for
cohorts. cognitive decline should not be eliminated, namely, mor-
Although women outperformed men on memory tality. Men have a lower life expectancy than women be-
function and information processing speed, their cogni- cause they die younger due to stroke or other cardiovas-
tive functions declined stronger with aging. Among the cular diseases. Therefore, it can also be argued that wom-
older population of Cache Country (USA), the incidence en have a higher risk of dementia due to a lower risk to
of AD in women was similar before the age of 85 years, die from other diseases earlier in life. This hypothesis was
but greater than men after the age of 85 years [29]. In Eu- verified in more recent analyses on the Framingham
ropean populations, the prevalence of dementia in men study that concluded that selective survival of men with a
and women start to diverge between the age of 80 and 85 healthier cardiovascular risk profile and hence lower pro-
years, to the detriment of women [30]. This age range is pensity to dementia might partly explain the higher life-
about equal to the age range where cognitive functions of time risk of dementia/Alzheimer’s disease in women [34].
men and women start to cross in our analyses (informa- Also in our study, the percentage of (surviving) women
tion processing speed and [in DCS] global cognitive func- increased with longer follow-up, especially at higher ages
tion). This disappearance of the female advantage may in the LASA. Our analyses excluding participants over 80
partly explain the female disadvantage in dementia inci- years generated similar, although somewhat larger sex
dence at very old age. differences in cognitive aging, indicating that healthy sur-
The difference in cognitive functions between men vivors narrow the sex differences in cognitive aging.
and women became smaller at higher age, since cognitive Probably, especially more healthy men with better cogni-
decline was larger in women than in men. It was not in tive health survive and thereby close the cognitive gap
the scope of the present study to elucidate on determi- between men and women at higher age, when less healthy
nants of these differences. Several mechanisms have been men have dropped out due to illness or death. Other ex-
postulated to explain sex differences in cognitive decline planations for sex differences in cognitive decline may be
and/or dementia prevalence: first, estrogen loss in post- found in differences in lifestyle or differences in cardio-
menopausal women may explain a stronger cognitive de- vascular risk factors. Another reason could also be a sta-
cline with aging, especially in verbal tasks [31]. Typically, tistical one: since women scored higher at baseline, they
we observed a higher rate of cognitive decline in women have a higher chance to decline stronger over time than
for memory and processing speed (and the sex differenc- men (regression to the mean).
es in these declines were larger after the age of 55 years in The LASA and DCS are both large population-based
DCS), but not for flexibility and fluid intelligence. There- cohort studies with a long follow-up period for cognitive
fore, estrogen loss may partly explain the observed sex function. Therefore, these cohorts provide data over a
differences in cognitive decline in the present study. large age range over different birth cohorts. Sensitive cog-
However, one may speculate that women in more recent nitive test batteries produced large ranges of cognitive
birth cohort may more often use estrogens as hormone performance, already from middle age. As all (long-last-
therapy around menopause, but sex differences in cogni- ing) cohort studies, both the LASA and DCS suffer from
tive decline between birth cohorts did not reflect such a (selective) attrition, resulting in relatively healthy co-
secular trend. Second, a greater cognitive reserve in men horts. Nevertheless, also these healthy cohorts showed
could explain higher dementia prevalence at similar age clear cognitive decline with aging, as well as differences
in women [31]. Individuals with greater cognitive reserve, between men and women, birth cohorts, and educational
as reflected in years of education, are better able to cope levels. Persons who dropped out had on average lower
with AD brain pathology without observable deficits in memory function than persons who did not drop out, but
cognition [32]. This may have been the case in our study. differences between men and women in memory func-

Sex Differences in Cognitive Aging Gerontology 2022;68:999–1009 1007

DOI: 10.1159/000520318
tion and memory decline were not different in the DCS the Medical Ethics Committee of the University of Utrecht (WMO
(data not shown). In the LASA, sex differences in memo- protocol number 07/233 and METC protocol number 07-233/O).
All participants gave written informed consent.
ry function, but not memory decline, were larger in the
group who remained >10 years in the study than in per-
sons who dropped out earlier. This means that sex differ-
ences observed in the LASA may be overestimated due to Conflict of Interest Statement
selective attrition. However, based on our sensitivity None of the authors reported a potential conflict of interest.
analyses on attrition in the DCS, attrition did not affect
estimates on sex differences. Since sex differences in the
LASA and DCS were comparable, we believe a potential
overestimation in LASA was only marginal. Another Funding Sources
point is the large difference in the educational level be- This work was supported by the Netherlands Organization for
tween the DCS and LASA. In the early 20th century, girls, Health Research and Development (ZonMw 849200005). The
regardless of their cognitive abilities, often did not attend Longitudinal Aging Study Amsterdam is supported by a grant
school longer than legally required: up to and including from the Netherlands Ministry of Health, Welfare, and Sport, Di-
primary school, as was especially seen in our earlier co- rectorate of Long-Term Care. The data collection (in 2012–2013
and 2,013–2,014) was financially supported by the Netherlands
horts. With women’s emancipation, it also became more Organization for Scientific Research (NWO) in the framework of
common for girls to attend school longer and the level of the project “New Cohorts of young old in the 21st century” (file
schooling better matches their cognitive abilities [11]. number 480-10-014). The Doetinchem Cohort Study is financially
This trend was confirmed in our population (online sup- supported by the Dutch Ministry of Health, Welfare, and Sport,
pl. Table 3). In both our cohorts, both the decline in per- and the National Institute for Public Health and the Environment.
The data up to and including 1997 were additionally financially
centage of low-educated participants and the increase in supported by the Europe against Cancer program of the European
percentage of high-educated participants were larger in Commission (DG SANCO). The funders had no role in the design
women than in men. Thus, as a proxy for cognition, edu- of the study, the collection and analysis of data, and the decision
cation is better suited in later cohorts than in earlier ones. to publish or the preparation of the manuscript.
In conclusion, at middle age, women had a better cog-
nitive function than men, but most sex differences in cog-
nitive function narrowed with aging. The observation Author Contributions
that women decline faster in memory and information
processing speed may reflect the sex differences in de- All authors met the criteria for authorship of the International
Committee of Medical Journal Editors. Specific contributions:
mentia prevalence as observed at old age. Female advan- S.H.O., L.S.S., and H.A.H.W. contributed to conception and acqui-
tage was larger in more recent birth cohorts. This may sition of the work; A.C.J.N. contributed to design, analysis, and
have implications for future sex differences in dementia drafting; and all authors contributed to interpretation of data, re-
incidence and prevalence. vising, and final approval of the manuscript.

Acknowledgments Data Availability Statement

The authors thank the respondents, epidemiologists, and field- The data of the Doetinchem Cohort Study pertaining to this
workers of the Longitudinal Aging Study Amsterdam and the Mu- manuscript are available upon request. We are unable to place data
nicipal Health Service in Doetinchem for their contribution to the in a public repository due to legal and ethical constraints. The par-
data collection for this study. ticipants’ informed consent did not include consent to public
availability of the data. However, the data are available upon re-
quest. We advise researchers to contact the corresponding author,
who will contact the scientific committee of the Doetinchem Co-
Statement of Ethics hort Study.
Both the DCS and LASA have been conducted in line with the
Helsinki Declaration. The LASA has been approved by the Medical
Ethics Committee of the VU University medical center (reference
numbers 92/138, 2002/141 and 2012/361) and the DCS by the ex-
ternal Medical Ethics Committee of the Netherlands Organization
for Applied Scientific Research (MEC codes 93/01 and 98/01) and

1008 Gerontology 2022;68:999–1009 Nooyens et al.

DOI: 10.1159/000520318
 References
 1 CBS. One in ten deaths attributed to demen- Amsterdam: cohort update 2016 and major 24 Van der Elst W, van Boxtel MP, van Breuke-
tia: Statistics Netherlands (CBS); 2017 [up- findings. Eur J Epidemiol. 2016;31(9):927–45. len GJ, Jolles J. Normative data for the animal,
dated 2017 Jul 19]. Available from: https: // 14 Hoogendijk EO, Deeg DJH, de Breij S, Klok- profession and letter M naming verbal fluen-
www.cbs.nl/en-gb/news/2017/29/one-in- gieters SS, Kok AAL, Stringa N, et al. The Lon- cy tests for Dutch speaking participants and
ten-deaths-attributed-to-dementia. gitudinal Aging Study Amsterdam: cohort the effects of age, education, and sex. J Int
  2 Prince MJ, Wimo A, Guerchet M, Ali G-C, update 2019 and additional data collections. Neuropsychol Soc. 2006;12(1):80–9.
Wu Y-T, Prina M. World Alzheimer report Eur J Epidemiol. 2019;35:61–74. 25 Nooyens AC, Bueno-de-Mesquita HB, van
2015. 2015. 15 Verschuren WM, Blokstra A, Picavet HS, Boxtel MP, van Gelder BM, Verhagen H, Ver-
  3 Rajan KB, Wilson RS, Weuve J, Barnes LL, Ev- Smit HA. Cohort profile: the Doetinchem Co- schuren WM. Fruit and vegetable intake and
ans DA. Cognitive impairment 18 years be- hort Study. Int J Epidemiol. 2008;37(6):1236– cognitive decline in middle-aged men and
fore clinical diagnosis of Alzheimer disease 41. women: the Doetinchem Cohort Study. Br J
dementia. Neurology. 2015;85(10):898–904. 16 Picavet HSJ, Blokstra A, Spijkerman AMW, Nutr. 2011;106(5):752–61.
  4 Hyde JS. Sex and cognition: gender and cog- Verschuren WMM. Cohort profile update: 26 Nooyens ACJ, Yildiz B, Hendriks LG, Bas S,
nitive functions. Curr Opin Neurobiol. 2016; the Doetinchem Cohort Study 1987–2017: van Boxtel MP, Picavet HSJ, et al. Adherence
38:53–6. lifestyle, health and chronic diseases in a life to dietary guidelines and cognitive decline
  5 Ferreira L, Ferreira Santos-Galduróz R, Ferri course and ageing perspective. Int J Epidemi- from middle age: the Doetinchem Cohort
CP, Fernandes Galduróz JC. Rate of cognitive ol. 2017;46(6):1751–g. Study. Am J Clin Nutr. 2021;114(3):871–81.
decline in relation to sex after 60 years-of-age: 17 Saan RJ, Deelman BG. Nieuwe 15-woorden 27 Aartsen MJ, Martin M, Zimprich D. Longitu-
a systematic review. Geriatr Gerontol Int. test A en B (15WTA en 15WTB) [new version dinal Aging Study A. Gender differences in
2014;14(1):23–31. of 15 words test (15WTA and 15WTB)]. In: level and change in cognitive functioning. Re-
  6 Caselli RJ, Dueck AC, Locke DE, Baxter LC, Bauma AMJ, Lindeboom J, editors. Neuro- sults from the Longitudinal Aging Study Am-
Woodruff BK, Geda YE. Sex-based memory psychologische diagnostiek Handboek [Neu- sterdam. Gerontology. 2004;50(1):35–8.
advantages and cognitive aging: a challenge to ro-psychological diagnostics handbook]. 28 van den Kommer TN, Deeg DJH, van der Fli-
the cognitive reserve construct? J Int Neuro- Amsterdam: Swets & Zeitlinger; 1986. p. 13– er WM, Comijs HC. Time trend in persistent
psychol Soc. 2015;21(2):95–104. 28. cognitive decline: results from the Longitudi-
  7 McCarrey AC, An Y, Kitner-Triolo MH, Fer- 18 Piccinin AM, Rabbitt PM. Contribution of nal Aging Study Amsterdam. J Gerontol B
rucci L, Resnick SM. Sex differences in cogni- cognitive abilities to performance and im- Psychol Sci Soc Sci. 2018;73(Suppl l):S57–64.
tive trajectories in clinically normal older provement on a substitution coding task. Psy- 29 Miech RA, Breitner JC, Zandi PP, Khachatu-
adults. Psychol Aging. 2016;31(2):166–75. chol Aging. 1999;14(4):539–51. rian AS, Anthony JC, Mayer L. Incidence of
  8 Zaninotto P, Batty GD, Allerhand M, Deary 19 Raven J, Raven JC, Court JH. Raven manual: AD may decline in the early 90s for men, later
IJ. Cognitive function trajectories and their selection 1. General overview. Oxford: Psy- for women: the Cache County study. Neurol-
determinants in older people: 8 years of fol- chologist Press; 1995. ogy. 2002;58(2):209–18.
low-up in the English Longitudinal Study of 20 Folstein MF, Folstein SE, McHugh PR. “Mini- 30 Lobo A, Launer LJ, Fratiglioni L, Andersen K,
Ageing. J Epidemiol Community Health. mental state”. A practical method for grading Di Carlo A, Breteler MM, et al. Prevalence of
2018;72(8):685–94. the cognitive state of patients for the clinician. dementia and major subtypes in Europe: a
  9 Levine DA, Gross AL, Briceno EM, Tilton N, J Psychiatr Res. 1975;12(3):189–98. collaborative study of population-based co-
Giordani BJ, Sussman JB, et al. Sex differences 21 Van der Elst W, van Boxtel MP, van Breuke- horts. Neurologic Diseases in the Elderly Re-
in cognitive decline among US adults. JAMA len GJ, Jolles J. Rey’s verbal learning test: nor- search Group. Neurology. 2000; 54(11 Suppl
Netw Open. 2021;4(2):e210169. mative data for 1855 healthy participants aged 5):S4–9.
10 Miller DI, Halpern DF. The new science of 24–81 years and the influence of age, sex, edu- 31 Laws KR, Irvine K, Gale TM. Sex differences
cognitive sex differences. Trends Cogn Sci. cation, and mode of presentation. J Int Neu- in cognitive impairment in Alzheimer’s dis-
2014;18(1):37–45. ropsychol Soc. 2005;11(3):290–302. ease. World J Psychiatry. 2016;6(1):54–65.
11 Vermeulen M, Waslander S. The Dutch way 22 Van der Elst W, van Boxtel MP, van Breuke- 32 Roe CM, Xiong C, Miller JP, Morris JC. Edu-
in education. Chapter 6: excellence in eman- len GJ, Jolles J. The letter digit substitution cation and Alzheimer disease without demen-
cipation, a century-long search for balance. test: normative data for 1,858 healthy partici- tia: support for the cognitive reserve hypoth-
Helmond, The Netherlands: Uitgeverij OMJS; pants aged 24–81 from the Maastricht Aging esis. Neurology. 2007;68(3):223–8.
2017. Study (MAAS)–influence of age, education, 33 Seshadri S, Beiser A, Kelly-Hayes M, Kase CS,
12 Huisman M, Poppelaars J, van der Horst M, and sex. J Clin Exp Neuropsychol. 2006;28(6): Au R, Kannel WB, et al. The lifetime risk of
Beekman AT, Brug J, van Tilburg TG, et al. 998–1009. stroke: estimates from the Framingham
Cohort profile: the Longitudinal Aging Study 23 Van der Elst W, van Boxtel MP, van Breuke- Study. Stroke. 2006;37(2):345–50.
Amsterdam. Int J Epidemiol. 2011;40(4):868– len GJ, Jolles J. The Stroop color-word test: 34 Chene G, Beiser A, Au R, Preis SR, Wolf PA,
76. influence of age, sex, and education; and Dufouil C, et al. Gender and incidence of de-
13 Hoogendijk EO, Deeg DJ, Poppelaars J, van normative data for a large sample across the mentia in the Framingham Heart Study from
der Horst M, Broese van Groenou MI, Comi- adult age range. Assessment. 2006; 13(1): 62– mid-adult life. Alzheimers Dement. 2015;
js HC, et al. The Longitudinal Aging Study 79. 11(3):310–20.

Sex Differences in Cognitive Aging Gerontology 2022;68:999–1009 1009

DOI: 10.1159/000520318